AN EFFICIENT STRATEGY FOR ANALYSIS AND VERIFICATION IN RULE-BASED EXPERT-SYSTEM DEVELOPMENT

A novel strategy for analyzing knowledge bases is presented, which assists designers of rule-based systems in assuring correctness during development. By helping to pinpoint flaws during the design phase via interactive as well as automated evaluation, such a tool can significantly reduce verification and validation efforts in the system test phase. Integrated into the environment of nonmonotonic forward-chaining rule-based systems, this strategy examines several aspects of consistency and identifies relations between rules according to the corresponding data values. The method of breaking down the knowledge components offers expert system developers valuable insights into the internal relationships among rules while providing the means for efficient automated analysis as well. A visual representation is introduced which makes the interdependencies transparent.